1. Field of the Invention
The present invention relates to an AGC circuit for calculating the average received level per slot, obtaining the difference between the calculated level and a reference value, and feeding back the difference to an AGC amplifier, thereby correcting any variations in the received level.
2. Description of the Prior Art
In mobile communications, the received signal strength greatly varies (a maximum of 80 dB is expected) owing to the influences of the distance between a base station and a terminal, fading due to the movement of a terminal, shadowing due to obstacles such as buildings, and the like. To stably receive and demodulate/decode such signals, the received levels must be corrected by using an AGC (Automatic Gain Control) circuit so as to level the received baseband signals.
This AGC circuit has been generally used in TV receivers and radios. According to a portable telephone based on the GSM or IS-95 CDMA scheme, a signal is transmitted from a base station in units of slots, and the signal is received and demodulated in units of slots on the terminal side. In this scheme, AGC must be performed to make the gain of the AGC amplifier constant within one slot, i.e., to make the relative received signal strength constant within one slot. Such an AGC circuit cannot be realized by a conventional AGC circuit designed to perform only analog signal processing. To realize this circuit, a method including digital signal processing in FIG. 1 is required.
FIG. 1 shows a conventional AGC circuit having such an arrangement.
FIG. 1 is a block diagram showing the schematic arrangement of the AGC circuit.
Since this block diagram of FIG. 1 aims at explaining only AGC operation, other portions that are not directly associated with AGC are omitted.
In this circuit, a received signal is a QPSK-modulated wave and is demodulated into two baseband signals, i.e., an I (In-phase component) and Q (Quadrature component) signals.
The received signal is converted into an IF signal by a receiver 1. The IF signal is amplified or attenuated by an AGC amplifier 2 and demodulated into I and Q baseband signals by a quadrature demodulator 3. The I and Q signals are respectively converted into digital signals by 8-bit A/D converters 4 and 5 (not limited to 8 bits).
Assume that a CDMA scheme with a chip rate of 4.096 MHz is used as a modulation scheme, and each A/D converter samples at 16.384 MHz, which is a conversion rate four times the chip rate. Assume also that the slot length is 625 xcexcsec (i.e., 2,560 chips). In this case, therefore, each of the numbers of I and Q signal samples obtained amounts to 2560xc3x974=10240 in a 1-slot interval.
A calculator 6 of received level calculates the average received level in the 1-slot interval from the above digital I and Q signals, and outputs the calculation result as an 8-bit straight binary code.
The calculator 6 of received level calculates the average received level in the following manner (for example).
Since each of the I and Q signals is an 8-bit signal from a positive peak to negative peak, the absolute value of each signal is obtained first. The maximum absolute value is obtained when each signal is completely saturated to the positive and negative peaks to have a rectangular waveform. This value of each of the I and Q signals is xe2x80x9c01111111xe2x80x9d in binary notation (xe2x80x9c127xe2x80x9d in decimal notation).
A received amplitude A should be calculated by:
A={square root over (I2+Q2)}xe2x80x83xe2x80x83(1)
However, since it is difficult to implement this calculation by hardware, an approximate value Axe2x80x2 given by:                               A          xe2x80x2                =                              Max            ⁡                          (                                                |                  I                  |                                ,                                  |                  Q                  |                                            )                                +                                    Min              ⁡                              (                                                      |                    I                    |                                    ,                                      |                    Q                    |                                                  )                                      2                                              (        2        )            
is used. FIG. 2 shows how the amplitudes A and Axe2x80x2 differ from each other.
The circle in FIG. 2 (sequence 3) is the result obtained by plotting a vector (I, Q) when the amplitude A is normalized with 1. A plot of the amplitude Axe2x80x2 on each vector yields the graphic pattern of sequence 1.
As can be seen from FIG. 2, the value Axe2x80x2 is always slightly larger than the value A. The value Axe2x80x2 is larger than the value A by an average of about 1.087 times, i.e., about 0.723 dB. Such a difference poses no problem in the AGC circuit. When the average received level of many samples corresponding to one slot (4xc3x972560 samples=10240 samples) or more is to be obtained, we can assume that the obtained amplitude is always larger than the true amplitude by 0.723 dB.
A method of calculating the average received level in a 1-slot interval will be described next.
The average received level is obtained by dividing the sum of the values Axe2x80x2 corresponding to one slot by the number of samples. Although the number of samples corresponding to one slot is 4xc3x972560=10240, since it is difficult to divide by using this number, 213=8192, which is the nearest power of 2, is used.
That is, the sum is shifted by 13 bit positions to the right. In other words, the output from the calculator 6 of received level is the result obtained by adding up the approximate values Axe2x80x2 corresponding to one slot (10,240 samples) and shifting the sum by 13 bit positions to the right.
Since the maximum value of each of the I and Q signals is 127, a maximum value Amax obtained by the above calculation is expressed by:                               A          ⁢                      xe2x80x83                    ⁢          max                =                                            (                              127                +                                  127                  2                                            )                        xc3x97                          10240              ÷              8192                                ≈          237                                    (        3        )            
This value can be expressed by an 8-bit straight binary code.
As described above, the average received level is calculated by calculating an approximate value of the average of amplitudes.
To be exact, the received level must be determined by obtaining the average of received powers. It is, however, difficult to calculate an average received power, because no suitable approximation method is available. Under the circumstances, an amplitude average approximate value is unwillingly used. The difference between the received level values respectively obtained by using power averages and amplitude averages will be examined.
A received signal S from a base station has in-phase and quadrature components written as:
S=I(t)xc2x7cos(2xcfx80fct)xe2x88x92Q(t)xc2x7sin(2xcfx80fct)xe2x80x83xe2x80x83(4)
In CDMA, since I(t) and Q(t) represent the sums of many independent speech channels, interference waves, and noise, the central limiting theorem holds. It therefore follows from this that I(t) and Q(t) each exhibit a Guassian distribution. If quadrature components respectively have independent Guassian distributions in this manner, the amplitude distribution of the synthetic signal exhibits a Rayleigh distribution. If amplitudes R have a Rayleigh distribution, the probability density distribution is given by:                               P          ⁡                      (            R            )                          =                              R                          b              0                                ·                      exp            ⁡                          (                              -                                                      R                    2                                                        2                    ·                                          b                      0                                                                                  )                                                          (        5        )            
In equation (5), b0 is a positive constant. Omitting a description of intermediate calculation, the power average is expressed by:                                           R            2                    _                =                                            ∫              0              ∞                        ⁢                                          R                2                            ·                              R                                  b                  0                                            ·                              exp                ⁡                                  (                                      -                                                                  R                        2                                                                    2                        ·                                                  b                          0                                                                                                      )                                            ·                              ⅆ                R                                              =                      2            ⁢                          b              0                                                          (        6        )            
Omitting a description of intermediate calculation, the amplitude average is be expressed by:                               R          _                =                                            ∫              0              ∞                        ⁢                          R              ·                              R                                  b                  0                                            ·                              exp                ⁡                                  (                                      -                                                                  R                        2                                                                    2                        ·                                                  b                          0                                                                                                      )                                            ·                              ⅆ                R                                              =                                                                      b                  0                                ·                π                            2                                                          (        7        )            
The dB difference between the received levels obtained by using the power and amplitude averages is given by:                     d        =                              20            ·                          log              ⁡                              (                                                                                                    R                        2                                            _                                                                            R                    _                                                  )                                              =                                    (                                                                    2                    ·                                          b                      0                                                                                                                                                          b                        0                                            ·                      π                                        2                                                              )                        =                                          20                ·                                  log                  ⁡                                      (                                                                  4                        π                                                              )                                                              ≈                              1.05                ⁢                                  xe2x80x83                                ⁢                dB                                                                        (        8        )            
That is, the received level calculated by using the power averages is equivalent to the value obtained by adding 1.05 dB to the received level calculated by using amplitude averages.
A method of determining reference level as an AGC target will be described next.
For example, the peak factor of a single code of a CDMA received signal is about 6 dB. When many users, e.g., 32 users, use this device, the peak factor increases by 30 dB to become 36 dB. It is, however, inadvisable to waste the dynamic range for a peak that rarely appears. In practice, it is probably appropriate to set the peak factor to about 10 to 12 dB. It follows from the foregoing that a value smaller than the peak value by 18 dB as a whole is appropriate as a reference value for the average received level, estimating a tracking error in AGC of about 6 dB.
Consequently, a reference value Aref is given by:                     Aref        =                                            A              ⁢                              xe2x80x83                            ⁢              max                                      10                              18                20                                              =          29.6                                    (        9        )            
Since this value is close to 32, which is the fifth power of 2, 32 is used as a reference value for the sake of simplicity. The dB error obtained when 32 is used in place of 29.6 is about 0.68 dB. This value can almost cancel out 0.723 dB, the dB error with an approximation calculation.
Feedback amount for an AGC value is determined by comparing the average received level calculated by the calculator 6 of received level with the reference value. Assume that the gain of the AGC amplifier 2 increases as the set value in a D/A converter 13 increases. A linearizer 12 corrects nonlinearity of the control voltage/gain characteristics of the AGC amplifier 2. Since this circuit is not directly associated with the present invention, a description thereof will be omitted.
After the AGC value which is the result obtained by adding up feedback values through an accumulator (comprised of an adder 10 and a register 11) is corrected by the linearizer 12, the AGC amplifier 2 is controlled by the AGC control voltage obtained by converting the corrected value through the D/A converter 13.
If, therefore, the average received level is higher than the reference level, the feedback amount becomes a positive value, and vice versa. However, since the AGC value is expressed in dB, the difference between the average received level and the reference level cannot be directly used as a feedback value.
The feedback value must be a value expressed in dB on the basis of the value obtained by dividing the average received level by the reference level. Since it is difficult to implement this calculation by means of hardware, a table 7 of feedback value, which uses an 8-bit average received level as an input address, is used in an actual circuit.
Each output from the table 7 of feedback value represents the dB difference between the average received power and the reference value. The accumulator (comprised of the adder 10 and the register 11) adds up the products of such outputs and appropriate coefficients 9. The linearizer 12 controls the AGC amplifier 2 through the D/A converter 13 by using the sum. This control is performed in units of slots.
The following problem is posed in the conventional circuit described above.
The dynamic range of the quadrature demodulator 3 is limited, and the number of bits of each of the A/D converters 4 and 5 is limited (e.g., 8 bits). If, therefore, an input signal having a received level excessively higher than the reference level is received, the A/D conversion result is saturated, as shown in FIG. 3. For this reason, as shown in FIG. 4, with regard to an input higher than the reference level by 10-odd dB or more, a value (solid line) considerably smaller than the true value (dotted line) is used as feedback data. As a result, when the received signal has an excessively high level, the convergence speed of the AGC circuit decreases.
The present invention has been made in consideration of the above situation, and has as its object to provide an AGC circuit which exhibits a high convergence speed even when a signal having an excessively high level is abruptly received.
In order to achieve the above object, according to the main aspect of the present invention, there is provided an AGC circuit for correcting a variation in received level by feeding back feedback data based on a difference between an average received level and a reference value in units of slots to an AGC amplifier, wherein when the difference between the average received level and the reference value is not less than a predetermined value, a value of the feedback data is set to be larger than a normal value.
The present invention has the following auxiliary aspects associated with the above main aspect.
The feedback data in the main aspect is obtained by using a table of feedback value.
This circuit comprises an A/D converter for A/D-converting the received level, and the predetermined value in the main aspect depends on the number of output bits of the A/D converter.
The predetermined value associated with the difference between the average received level and the reference value is 10 dB.
In addition, there is provided a radio receiver or portable telephone comprising the AGC circuit defined in the main aspect or any one of the auxiliary aspects.
According to the AGC circuit of the present invention, even when a signal having an excessively high level is abruptly received, the convergence speed of AGC can be increased as compared with the prior art.
The above and many other objects, features and advantages of the present invention will become manifest to those skilled in the art upon making reference to the following detailed description and accompanying drawings in which preferred embodiments incorporating the principle of the present invention are shown by way of illustrative examples.